The present invention generally relates to reactive plasma sputter deposition techniques for forming and depositing insulating films on substrates and, more particularly, is concerned with a system and method for sputter deposition of an insulating material on a substrate.
Sputtering is a process wherein a target, usually a metal, is placed in position near a plasma (a cloud of ions and electrons in equal numbers), in a chamber in which most of the air has been withdrawn. Well-known conventional means are used to create the plasma. A negative voltage is produced on the target, or cathode, relative to a separate electrode called the anode by connecting the negative lead of a dc power supply to the target. The negative voltage on the target attracts the ions from the plasma, which are accelerated toward the target. Upon arrival the collision of the ions with the target physically knocks out target atoms. These target atoms travel from the target to a substrate placed nearby, which becomes coated with them. The expelled target atoms also coat every other surface in the system, as for the most part, they are neutral and there is no practical way to direct their path. When ions are withdrawn from the plasma, there immediately exists an excess of electrons in the plasma. These excess electrons are attracted to the positive lead of the dc power supply used to create the target voltage, which positive lead is connected to a separate electrode called the anode or alternatively to the chamber walls, either of which, in collecting the electrons, provide for plasma current flow and therefore may be considered as plasma current providing elements.
As described, this is a very common process for deposition of thin layers of metals. It is widely used in the processing of semiconductors, and in creating the reflecting layer on compact discs and CD-ROMS, active layers on hard discs for computer storage, and layers of metals for many other functional and decorative applications.
The process described above is called dc sputtering, and requires that the target (or cathode) be conducting, because the ions arriving at the target must be able to accept one or more electrons from the target to become neutral gas atoms again in order to prevent charging of the target surface, which would create a retarding potential which would stop the process very quickly. Insulators do not have free electrons available for this purpose, so that an insulating target material cannot be used. On the other hand, one can deposit layers of insulating material from a metallic target, by forming the insulator chemically through reaction with a reactive background gas. This is called reactive sputtering. For example, Al.sub.2 O.sub.3 and SiO.sub.2 can be created from aluminum and silicon targets, respectively, if oxygen gas is present in appropriate quantities in the background gas filling the chamber.
There is increasing commercial interest in processes involving deposition of such insulating films. This interest comes about at least in part because of the application of such processes to the deposition of wear resistant coatings; insulating films for microcircuits (including devices such as thin film heads) or electronic devices such as capacitors; sophisticated architectural glass coatings; coatings on polyester film for architectural glass laminates or oxygen barriers for food packaging; heat reflecting coatings for high efficiency lamps or induction furnace heat shields; deposition of barrier and functional layers for flat panel displays, including the ITO glass used in LCD displays; and myriad other similar functional applications. Added to this are the many reactive PVD processes used to create decorative effects on a wide variety of plastic, natural and artificial fiber, and metal substrates.
A problem occurs in these cases, however, when the reaction product is an electrical insulator. Since, as described above, the insulating film coats every surface in the chamber (which it will eventually do) then it will surely eventually coat the anode. As this happens the conduction path for the electrons is coated over, and the process cannot be sustained. This has been termed the "disappearing anode" problem. In the past the reactive process was run until this effect began to create serious problems, whereupon the system was opened to mechanically scrub off the offending insulating layer from the anode to create a new metallic surface. Thus continuous operation without this maintenance is not possible.
Another drawback related to the coating of the anode with an insulator is that this insulator will generally charge up as the electrons attempt to collect there. This charge can cause an electric field in the insulating film on the anode which may exceed the dielectric strength of the film material. When this occurs an arc may be formed and the energy in this arc may cause portions of the film to be ejected from the anode, creating particulates which can become included in the film growing on the substrate, causing defects which may be unacceptable in the final product.
Este, et al, in an article entitled "A Quasi-direct-current Sputtering Technique for the Deposition of Dielectrics at Enhanced Rates", published in J. Vac. Sci. Technol. A, vol. 6, No. 3 (May/June 1988), proposed an approach to the sputtering process which uses two targets alternately for deposition of dielectric or insulating films. The power supply, which in this case has an alternating potential output, is connected to the two targets so that they are driven alternatively positive and negative with respect to one another. This causes each to act as an anode for the other. If the reversal takes place often enough, only a very thin layer of insulator will be formed on the target acting as an anode, and this very thin layer can be sputtered away when it is that target's turn to be negative. This is possible because the insulator does not stop the sputtering process at once, but due to charging effects its presence will slow and eventually stop the process. If the layer is very thin it can be sputtered away before the process stops. The usual time for reversal is a few tens of microseconds, in order that there be too little time for a thick layer to form. See also the paper by Schiller et al entitled "Pulsed Magnetron Sputter Technology", published in the Proceedings of the 1993 International Conference on Metallurgical Coatings and Thin Films, Surf. Coat. Tech. Vol. 61, (1993) page 331, which covers a dual magnetron target approach similar to that of Este et al in that each of the targets acts within one cycle of the output of the power supply once as the cathode and once as the anode.
For the most part this has proved to be a successful approach to the problem of the "disappearing anode". It does have the disadvantage, however, of requiring two targets, which adds to the expense of the system and also complicates the maintenance. Also, it is difficult to retrofit this dual target process into existing sputtering systems because there often is no room for the second target.
A more serious drawback to the dual target approach is caused by the fact that appropriate design of the target assembly usually involves magnets to create a magnetic field above the target surface to enhance the plasma density. This magnetic field impedes the flow of electrons to the target. Thus an appropriate design for a cathode is generally not a good design for an anode, which calls for unimpeded collection of electrons from the plasma. In a sputtering system there is a potential difference between the plasma and the target, or cathode, which is called the "cathode fall". Similarly there is a potential difference between the plasma and the anode, generally much smaller, which is called the "anode fall". In a well designed system almost all of the voltage of the power supply appears as cathode fall and little appears as anode fall. In a typical case the cathode fall might be 600 volts and the anode fall less than 20 volts. With the dual target system, on the other hand, the anode fall is increased to much larger values, often as large as 50 to 100 volts. This creates two serious symptoms. First, the higher anode fall changes the plasma potential such that a higher energy substrate bombardment occurs. This can be beneficial, as some ion bombardment can help make the growing film more dense, but substrate bombardment also equates with substrate heating, and this can be a significant disadvantage if the substrate is composed of a heat sensitive material, such as plastic. Secondly, since the power supply current passes through the anode, the product of the anode fall and the current represents power loss in the anode. This in itself may be a problem, but even if the anode can withstand the heating effect, the power loss in the anode must necessarily subtract from the power of the power supply (that is the actual device or even any part of the circuitry involved in or which acts to facilitate some supply of power to the element involved) and thus reduce the power available to the cathode for sputtering purposes. Thus the deposition rate equivalent to each watt delivered by the power supply is reduced by the proportion of the anode fall to the cathode fall. In the example case above for dual cathode sputtering, the anode would receive 1/6 of the power (the current times 100 volts out of the 600 volts available from the power supply) and the cathode the balance of 5/6. This represents a loss of 16.7% of the potential sputtering power, while an anode fall of 10 volts would "steal" but 1.7% of the potential power to heat the anode. Of course, other losses may also exist, but a substantial anode fall can materially lower the deposition rate.
Consequently, a need exists for a different approach to overcome the above-described drawbacks of the prior art reactive sputtering processes. The present invention permits single target operation with a separate anode assembly with a small anode fall and therefore low substrate bombardment and good utilization of the sputtering power, without suffering from the problem of the "disappearing anode".